ml-notes

Technical blog and index of a self-directed learning path on machine learning, inference systems and information retrieval. Every project lives in its own repository and ends with a post here.

Raw benchmark data is kept in data/, so every chart can be regenerated from source. Published numbers hold for one machine: an RTX 3060 12 GB, a Ryzen 5 8500G and 32 GB of RAM.

The path

#ProjectStatusRepoPost
0Foundations: autograd and a Transformer from scratchplannedfoundations-
1Format and quantization benchmark on an RTX 3060plannedgpu-quant-bench-
2Vector database in Rust: HNSW and Product Quantizationplannedrustann-
3OCR from scratch: CRNN with CTC loss on synthetic dataplannedcrnn-ocr-
4Semantic code searchplannedcodesearch-
5Hand-written CUDA kernels and profilingplannedcuda-kernels-
6LoRA fine-tuning on a domain of my ownplannedlora-lab-
7Disk failure prediction from SMART data (Backblaze)plannedsmart-failure-
8DQN, PPO and SAC from scratch, plus a study on seed varianceplannedrl-from-scratch-
-Shared utilities. Created when needed, not before-mlkit-

Statuses: planned, in progress, done, abandoned. Abandoned projects stay in the table with the reason. A repository stays private until its post ships.